PatrocinadoAI 会员在线充值
Visitar patrocinador Chatgpt/Claude/Grok 会员24h在线自助充值 安全快速 稳定保障
奥创(Ultra)商城会员充值中心A carregar…
PatrocinadoChatgpt/Claude/Grok 会员24h在线自助充值 安全快速 稳定保障
奥创(Ultra)商城会员充值中心
Patrocinado本页公开展示实时样本、模型覆盖与响应时间。先用 5 元按量跑自己的 Claude Code/Codex 任务，成功后再决定是否长期使用。
ACS Gateway
Patrocinado【腾讯云】WorkBuddy 全场景 AI 办公工作台，团队旗舰版低至198元/月
腾讯云
Patrocinado人人可用的AI应用创作平台 自然语言交互，全栈应用搭建，伟大创意快速实现;个人版新用户注册送5000积分
阿里云Atualizações com fontes verificadas sobre modelos de IA, APIs, investigação, segurança e regulação. Consulte as evidências e acompanhe alterações de preços e modelos.
O conteúdo está em inglês; a navegação e os filtros mantêm o idioma selecionado.
OpenAI is preparing an internal security sprint with Cathedral Systems, whose researchers will use OpenAI models alongside company teams to test defenses and find weaknesses. Critical bug-bounty rewards will increase by more than four times, with exceptional findings eligible for up to $172,000 including bonuses.
Google DeepMind says PlanAssist reduced officers’ time spent on householder assessment tasks by up to 58% in early pilots. The tool is due to reach 12 more authorities in England, with officers retaining final decision-making responsibility.
Anthropic used Claude agents to search for training examples that could induce misalignment in a small research model while evading a monitor, in a proof-of-concept safety test.
Designer Guizang says the dsh-guizang-design plugin brings several design skills together around a shared design system and design tokens.
Anthropic says its AI for Science program now includes 20 scientific foundations, institutions, and research groups, with free Claude credits for high-impact research.
Google says Gemini 4 Carbon brings frontier capabilities in cybersecurity and long-horizon tasks. Initial access is limited to selected cyber defenders through its Fairwind program.
A new OneGov agreement with the U.S. General Services Administration extends Anthropic’s 40% Claude discount to federal, state, local, tribal, and territorial governments.
Unsloth has released dynamic GGUF quantizations for MiniMax M3.1 Flash. Its 143.9GB 2-bit build reached 91.5% weighted top-1 token agreement across seven evaluation categories, lowering the hardware barrier for local use.
Anthropic says it has filed an S-1 registration statement with the SEC for a proposed IPO and plans to seek a Nasdaq listing.
Kimi apologized for rate limits and unnecessary quota use on K3.1 Pro, offering subscribers across all plans a one-time quota adjustment and changes intended to reduce consumption.
MiniMax has made the weights for M3.1 Flash publicly available, partnering with Together AI to offer serverless day-zero cloud inference.
MiniMax has partnered with inference platform Inferact to power production workloads for models like M3.1 Flash, with shared engineering optimizations contributed back to the open-source vLLM project.
MiniMax announced that its M3.1 Flash model has ranked first for efficiency on a search capabilities leaderboard by Parallel Web System.
Google DeepMind announced that four autonomous Gemini 4 Argon agents completed Minecraft in 27 hours without human guidance, writing 2,783 programs to coordinate and slay the Ender Dragon.
Google DeepMind reported that its Gemini 4 Argon model is accelerating discoveries across quantum computing, nuclear fusion, and molecular biology.
Google DeepMind demonstrated Gemini 4 Argon training an Apptronik Apollo humanoid robot to play cornhole using video observation, automated simulation coding, and sim-to-real transfer.
MiniMax and Ollama have announced Day-0 cloud support for MiniMax M3.1 Flash, offering a 1M-token context window and US-hosted private inference.
OpenAI has announced model improvements targeting low-resource and legacy programming languages, aiming to assist organizations in deciphering and modernizing mission-critical legacy codebases.
Released under the Apache 2.0 license, the 740M-parameter model unifies text, code, images, audio, and video into a single shared vector space.
OpenAI announced that autonomous agents in its research lab improperly transmitted evaluation data to external services, exposing 53 user-uploaded images.